A study by Next AI Labs has identified a critical barrier in the development of multi-agent systems: more than half of established external skills remain inaccessible to agents due to the lack of standardized metadata.
What Happened
According to data from Next AI Labs, there is a sharp gap in tool discoverability within multi-agent systems. While custom tools provide 100% visibility, external skills (e.g., from Trail of Bits or OWASP) are discovered in only 47% of cases. The problem lies in the absence of semantic descriptions that are understandable to agent search mechanisms.
Context
The scalability of multi-agent systems (MAS) today is limited not by computational power, but by the ability of agents to find the necessary tools. To solve this problem, the implementation of standardized YAML metadata with triggers is proposed, which would allow discovery through five levels: protocols, trigger matching, registries, Semantic Swarm Search, and registry semantics.
Why It Matters for the Industry
For the industry, this signifies a shift from developing purely functional tools to designing systems with "agentic discoverability" in mind. Without the implementation of unified tool schema standards and the creation of specialized registries, organizations will face redundant functional duplication and an inability to effectively utilize existing security and automation libraries.
Why It Matters for Users
AI solution developers and integrators need to implement strict semantic tagging schemes as early as the prototyping stage. Ignoring metadata during tool development leads to their "isolation," turning them into useless "dead weight" within skill libraries that agents simply cannot find or apply.
What Is Not Yet Known / Limitations
While technical specialists focus on architectural search methods, for the business segment (SaaS and product builders), the key question remains market scalability and the economic efficiency of using existing libraries.
Sources
Author
Look at AI, Editorial Staff